In industrial processes, reliable measurement systems are fundamental for effective quality control and decision-making. This study addresses the challenge of integrating decision-maker customer preferences into multivariate measurement system analysis, focusing on repeatability and reproducibility indices (%R&R), especially in contexts where correlated quality characteristics are involved. Traditional approaches in Measurement System Analysis often neglect the varying importance of quality characteristics or their interdependencies, which can hinder accurate decision-making. To overcome these limitations, the article proposes a novel methodology based on Weighted Principal Components and prioritization of quality characteristics according to their significance. Through numerical simulations and real data from an industrial stainless steel cladding process, the proposed method is validated across 18 scenarios, highlighting how assigning varying weights to characteristics impacts multivariate R&R indices. The findings reveal that prioritization alters the system's classification, aligning it with the most significant variables while also capturing correlations among characteristics. When 90 % of the weight was assigned to a single characteristic, the multivariate R&R index closely followed its univariate counterpart, whereas with 47.5 % distributed between two characteristics, the multivariate index assumed intermediate values. Furthermore, by adjusting priorities in a real industrial measurement system, the proposed method significantly enhanced measurement accuracy, reducing the multivariate %R&R index from 8.98 % (marginal classification) to 0.76 % (acceptable classification)—a 91.5 % reduction in measurement system variability.
扫码关注我们
求助内容:
应助结果提醒方式:
